Package opennlp.spellcheck.distance

Class DamerauOSADistance

java.lang.Object

opennlp.spellcheck.distance.DamerauOSADistance

All Implemented Interfaces:

EditDistance

public final class DamerauOSADistance
extends Object
implements EditDistance

Optimal String Alignment (restricted Damerau-Levenshtein) edit distance.

Counts insertions, deletions and substitutions, plus transpositions of two adjacent symbols, each with a unit cost. As an optimal-string-alignment metric, a given substring may not be edited more than once, which is the variant used by the SymSpell reference implementation.

This is the default edit distance for the engine. It is Unicode-aware: comparison happens on Unicode code points, so characters outside the Basic Multilingual Plane (e.g. many emoji) are treated as single symbols.

Instances are immutable and thread-safe. A bounded computation with early exit is provided through distance(CharSequence, CharSequence, int).

Field Summary

Fields

Modifier and Type	Field	Description
static final DamerauOSADistance	INSTANCE	Shared, stateless instance.

Constructor Summary

Constructors

Constructor	Description
DamerauOSADistance()

Method Summary

Modifier and Type	Method	Description
int	distance(CharSequence a, CharSequence b, int max)	Computes the edit distance between a and b, giving up early once it is certain the distance exceeds max.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Details

INSTANCE

public static final DamerauOSADistance INSTANCE

Shared, stateless instance.

Constructor Details

DamerauOSADistance

public DamerauOSADistance()

Method Details

distance

public int distance(CharSequence a,
 CharSequence b,
 int max)

Description copied from interface: EditDistance

Computes the edit distance between a and b, giving up early once it is certain the distance exceeds max.

A max of 0 is permitted and meaningful: it asks only whether the two sequences are equal, returning 0 when they are and -1 as soon as any difference is found. It is the natural lower bound of the contract, not a degenerate case.

Specified by:

distance in interface EditDistance

Parameters:

a - the first sequence; must not be null

b - the second sequence; must not be null

max - the maximum acceptable distance; must not be negative (>= 0)

Returns:

the edit distance, or -1 if it is strictly greater than max